Numerical methods are a preferred choice for predicting the behaviour of coupled THM processes, the technologies and methods for improving their accuracy have received a considerable amount of attention in the literature (Jing, 2003, Jing et al., 1993, Manteufel et al., 1993, Noorishad et al., 1984, Stephansson et al., 2004). Coupled processes in geo-systems involve multiple classes of physical and chemical processes into complex configurations. They often result in non-linear behaviours that require specifically developed numerical methods as it is not sufficient to model each process separately. THM processes are of utmost importance for GDF performance assessment, along with chemical processes they determine the long-term behaviour of the three-barriers system (Manepally et al., 2011, Stephansson et al., 2004, Tsang et al., 2004). Thus THM coupled numerical tools are in high demand to support the safety case of GDFs for regulatory approval. For the past four decades, several international projects led the research on advanced numerical methods for THM modelling such as the ongoing DECOVALEX project (since 1992, decovalex.org), the Yucca Mountain repository project (Manteufel et al., 1993) or the EU-funded THERESA project (THERESA, 2009). They brought together technologies from both the research and private sector in an e↵ort to understand and predict the processes at play, establish a